1 repo scan
Starter License
Planned regular €199Save €50
€149
Protect everyday AI use and your first repository.
Public proof ledger
Inspect the evidence before buying CapitalGuard.
This page collects the real public surfaces behind the product: package scope, checkout readiness, delivery vault, scanner assets, demos, release notes, and claim boundaries.
3
Public license products
10
AI-agent exposure modules
5
Pro scan credits
1 year
Updates included
Checkout evidence
Three fixed license packages are ready to buy.
Each product page explains scope before payment. Customers continue to hosted checkout from the selected product page.
5 repo scans
Pro License
Planned regular €699Save €200
€499
Make AI-agent exposure visible, governable, and provable.
25 repo scans
Agency License
Planned regular €2,799Save €800
€1,999
Turn AI-agent security into a client-ready delivery system.
Live
Hosted checkout
Starter, Pro, and Agency route customers to Stripe-hosted payment pages.
Live
Delivery vault
Every license has a post-payment activation, scan intake, customer kit, and access recovery path.
Available
Scanner assets
The customer kit includes a guarded installer, policy template, safe-use guide, and delivery checklist.
Beta
Runtime gate
Enforced Mode removes ambient host authority, Delegation Firewall attenuates child authority, and MCP Integrity Gate pins external tool identity and schemas.
Public
Verification surface
Paid customers can verify restrained badge claims after authorization and operator approval.
Product evidence
Customers can inspect the workflow before checkout.
The demos are not customer results. They are product surfaces that show the report style, delivery path, scanner intake, policy generator, dashboard preview, and verification flow.
Open Demo LibraryContained mode
Enforced Mode Runtime Gate
Inspect the filtered workspace, digest-pinned OCI boundary, authenticated broker, replay protection, and exact limits of the contained session.
Signature
CapitalGuard AccessGraph
Explore the action-to-asset map, required controls, and privacy-safe Guardprint that distinguish CapitalGuard from a generic code scanner.
Playbooks
Prevention Playbooks
Build package-specific action packets for secret exposure, prompt injection, workflow risk, and customer data context.
Guardrails
Guardrail Installer
Inspect the local installer and policy planner for blocked paths, protected files, approval gates, and redaction rules.
Scanner
Repo Exposure Scanner
Run a browser-side triage scan on pasted file trees and redacted config signals before buying or submitting licensed scope.
Service
License Console
Open the service workflow customers use for activation, scan inventory, authorized intake, reports, downloads, and monitoring readiness.
Checklist
AI-Agent Checklist
Use the public checklist to identify missing repository controls and route customers to the right license path.
Report
Sample Risk Report
Inspect the kind of finding evidence a customer receives: severity, affected path, business impact, confidence, and preventive control.
Delivery
Customer Delivery Vault
See the post-checkout flow that gives customers activation, scan intake, access recovery, and customer-kit downloads.
Scanner
Scanner Intake
Review the authorization-first scan intake used before any repository scope is handled.
Policy
Policy Generator
Generate practical policy starters for AI coding tools with repository access.
Dashboard
Dashboard Preview
View the operational dashboard concept for risk score, dangerous files, permissions, and policy status.
Trust
Public Badge Verification
Check the public verification surface for teams that need a restrained proof point after paid review.
Downloadable assets
The customer kit has tangible security deliverables.
Repository Exposure Evidence BriefConcise release summary for the six-case synthetic repository exposure benchmark, with reproducible artifacts and explicit limitations.Repository Exposure BenchmarkSix disclosed synthetic exposure classes, case-level results, fixture pack, release hashes, and stated limits.Agent Runtime Gate BenchmarkTwenty-two reproducible containment and broker cases with a fixture pack, release hashes, and explicit limitations.Enforced ModeFiltered workspace, hardened OCI session, authenticated broker, one-time approval leases, and tamper-evident events.Agent Delegation FirewallPublic target-binding, authority-attenuation, no-fan-out, expiry, depth, replay, and residual-boundary contract.Agent Delegation Boundary BenchmarkTwelve fixed synthetic cases for signed chains, authority attenuation, actor and policy binding, depth, and atomic single use.MCP Integrity GatePublic server-identity, tool-inventory, schema-drift, side-effect approval, replay, output-quarantine, and residual-boundary contract.MCP Tool Integrity BenchmarkPublisher-run synthetic evidence for identity, tool-description, input-schema, approval, replay, quarantine, and size-boundary changes.MCP Tool Review GuideA cited five-step workflow and blank review register for server identity, tool contracts, authority, approvals, and evidence.Local Guardrail InstallerVerified customer download for local repository policy and AI-agent precautions.Safe-Use GuidePractical public guidance with the full operating file reserved for verified customers.Policy FileInteractive policy rules for protected paths, approvals, and redaction.License Kit ArchiveProtected customer deliverables unlocked with an active license and purchase email.
Release ledger
Shipping history is visible instead of vague hype.
Product releases are listed with dates and concrete surfaces so customers can see what exists now.
MCP Integrity Gate 0.1
Signed tool inventories and output quarantine
July 18, 2026
Every approved MCP server is locked to one measured identity digest, protocol version, exact tool list, descriptions, input schemas, and output schemas.
Mutating and unknown tools require an already-verified one-time gateway approval; every request is reserved before invocation so sequential and concurrent replay fail closed.
Untrusted provider output is quarantined outside agent context by default, while public schemas, vectors, release digests, and residual boundaries make the control inspectable.
Agent Delegation Firewall 0.1
Single-use authority chains for multi-agent work
July 18, 2026
Every signed delegation binds one target agent, task, policy, root session, capability subset, expiry, and maximum chain depth.
A parent envelope can be spent on one child or one terminal request, never both; sequential and concurrent replay fail closed in a lock-protected local ledger.
Public schemas, synthetic signatures, threat model, support matrix, and release digests state the exact control and the live-adapter boundary.
Enforced Mode 0.3
Unified runtime trust chain for contained AI-agent sessions
July 19, 2026
The licensed launcher builds a policy-filtered read-only workspace and starts a digest-pinned OCI image without direct network, host-source, Docker-socket, or ambient-secret access.
Every dynamic request is task-bound, signed as model-originated context, derived as untrusted, checked against sequence history, and then evaluated by the default-deny gateway.
Separate Context Provenance and Runtime Sequence authorities, external security-state roots, and authenticated pending state make replay, rollback, and uncertain crash recovery fail closed.
The 22-case publisher-operated benchmark and signed open challenge publish the exact boundary, evidence, provenance, and limitations without unsupported competitor scores.
Agent Action Gateway 0.1
Routed operation enforcement with replay-protected approvals
July 16, 2026
The licensed gateway enforces supported file, command, and metadata-only network operations submitted through its local adapters.
Approval leases are signed, short-lived, bound to exact request and policy digests, and atomically limited to one use.
Every decision appends a privacy-safe tamper-evident event and states that operations bypassing the gateway remain outside its control.
Claim boundaries
Clear boundaries make the product easier to trust.
No repository is scanned until ownership or written authorization is confirmed.
Secrets and sensitive values are redacted in customer-facing reports and demos.
CapitalGuard detects exposure and recommends precautions; it does not guarantee breach prevention.
Policy artifacts return declared decisions; they do not intercept runtime operations or prove enforcement.
Delegation Firewall authorizes a signed handoff but does not start the target process or replace a required approval lease.
MCP Integrity Gate controls routed calls only; direct calls, dishonest identity measurement, approved-tool vulnerabilities, and host compromise remain outside its authority.
Public badge claims are issued only after payment, authorization, scan completion, and operator approval.
Published proof uses product evidence, release notes, and checkout status instead of unsupported testimonials.
Ready to buy